[{"data":1,"prerenderedAt":185},["ShallowReactive",2],{"content-query-pLMSRQVEPl":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":179,"_id":180,"_source":181,"_file":182,"_stem":183,"_extension":184},"/news/zh/3600","zh",false,"","昇思MindSpore全流程支持DeepSeek V3满血版训推","昇思MindSpore开源社区上线DeepSeek V3微调样例","2025-02-11","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/02/11/133430cd38a345f38d055a6190291bef.png","news",{"type":14,"children":15,"toc":176},"root",[16,24,30,39,93,101,112,120,128,136,144,156,164],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"昇思mindspore全流程支持deepseek-v3满血版训推",[22],{"type":23,"value":8},"text",{"type":17,"tag":25,"props":26,"children":27},"p",{},[28],{"type":23,"value":29},"继昇思MindSpore支持在昇腾集群上训练和推理DeepSeek V3 671B，DeepSeek V3昇思MindSpore版本新增微调能力同步上线，至此昇思MindSpore可全流程端到端支持DeepSeek V3训推。",{"type":17,"tag":25,"props":31,"children":32},{},[33],{"type":17,"tag":34,"props":35,"children":36},"strong",{},[37],{"type":23,"value":38},"开源链接",{"type":17,"tag":40,"props":41,"children":42},"ul",{},[43,54,59,70,79,84],{"type":17,"tag":44,"props":45,"children":46},"li",{},[47,52],{"type":17,"tag":34,"props":48,"children":49},{},[50],{"type":23,"value":51},"·",{"type":23,"value":53}," 昇思MindSpore开源社区",{"type":17,"tag":44,"props":55,"children":56},{},[57],{"type":23,"value":58},"DeepSeek V3训练微调代码：",{"type":17,"tag":44,"props":60,"children":61},{},[62],{"type":17,"tag":63,"props":64,"children":68},"a",{"href":65,"rel":66},"https://gitee.com/mindspore/mindformers/tree/dev/research/deepseek3",[67],"nofollow",[69],{"type":23,"value":65},{"type":17,"tag":44,"props":71,"children":72},{},[73,77],{"type":17,"tag":34,"props":74,"children":75},{},[76],{"type":23,"value":51},{"type":23,"value":78}," 魔乐社区",{"type":17,"tag":44,"props":80,"children":81},{},[82],{"type":23,"value":83},"DeepSeek V3推理代码：",{"type":17,"tag":44,"props":85,"children":86},{},[87],{"type":17,"tag":63,"props":88,"children":91},{"href":89,"rel":90},"https://modelers.cn/models/MindSpore-Lab/DeepSeek-V3",[67],[92],{"type":23,"value":89},{"type":17,"tag":25,"props":94,"children":95},{},[96],{"type":17,"tag":34,"props":97,"children":98},{},[99],{"type":23,"value":100},"链接中包含完整的手把手教程，助力开发者开箱即用",{"type":17,"tag":25,"props":102,"children":103},{},[104],{"type":17,"tag":34,"props":105,"children":106},{},[107],{"type":17,"tag":34,"props":108,"children":109},{},[110],{"type":23,"value":111},"DeepSeek V3昇思微调能力上线",{"type":17,"tag":25,"props":113,"children":114},{},[115],{"type":17,"tag":34,"props":116,"children":117},{},[118],{"type":23,"value":119},"MindSpore Transformers支持对DeepSeek V3进行全参微调。通过以下步骤，即可在单台Atlas 800T A2（64G）上快速拉起微调流程。",{"type":17,"tag":25,"props":121,"children":122},{},[123],{"type":17,"tag":34,"props":124,"children":125},{},[126],{"type":23,"value":127},"① 环境准备→②数据集准备→③模型权重准备→④修改配置→⑤拉起任务",{"type":17,"tag":25,"props":129,"children":130},{},[131],{"type":17,"tag":34,"props":132,"children":133},{},[134],{"type":23,"value":135},"在使用模型中，有任何疑问和建议，均可通过社区进行反馈。",{"type":17,"tag":25,"props":137,"children":138},{},[139],{"type":17,"tag":34,"props":140,"children":141},{},[142],{"type":23,"value":143},"昇思MindSpore开源社区DeepSeek V3使用问题讨论issue：",{"type":17,"tag":25,"props":145,"children":146},{},[147],{"type":17,"tag":34,"props":148,"children":149},{},[150],{"type":17,"tag":63,"props":151,"children":154},{"href":152,"rel":153},"https://gitee.com/mindspore/mindformers/issues/IBL0X5?from=project-issue",[67],[155],{"type":23,"value":152},{"type":17,"tag":25,"props":157,"children":158},{},[159],{"type":17,"tag":34,"props":160,"children":161},{},[162],{"type":23,"value":163},"昇腾社区昇思论坛DeepSeek V3使用问题讨论帖：",{"type":17,"tag":25,"props":165,"children":166},{},[167],{"type":17,"tag":34,"props":168,"children":169},{},[170],{"type":17,"tag":63,"props":171,"children":174},{"href":172,"rel":173},"https://www.hiascend.com/forum/thread-02112174450796469017-1-1.html",[67],[175],{"type":23,"value":172},{"title":7,"searchDepth":177,"depth":177,"links":178},4,[],"markdown","content:news:zh:3600.md","content","news/zh/3600.md","news/zh/3600","md",1776506085953]